Cancer Causes & Control

, Volume 17, Issue 1, pp 103–107

Ultraviolet Radiation at Places of Residence and the Development of Melanocytic Nevi in Children (Australia)


    • Cancer Epidemiology CentreThe Cancer Council Victoria
    • School of Population HealthThe University of Melbourne
  • Elizabeth Milne
    • Telethon Institute for Child Health Research, Centre for Child Health ResearchThe University of Western Australia
  • Julie A. Simpson
    • Cancer Epidemiology CentreThe Cancer Council Victoria
    • School of Population HealthThe University of Melbourne
Original Paper

DOI: 10.1007/s10552-005-0425-0

Cite this article as:
English, D.R., Milne, E. & Simpson, J.A. Cancer Causes Control (2006) 17: 103. doi:10.1007/s10552-005-0425-0


Objective To investigate the relationship between ambient ultraviolet radiation (UV) exposure and number of melanocytic nevi in children.

Methods A longitudinal study of nevi was conducted in 1614 children in Perth, Australia. Children had nevi counted on the back, face and arms at ages 6, 10, and 12 years. Erythemally effective UV irradiance was used to estimate ambient exposure from their places of residence before entry to the study at age 6 years. Data on UV radiation were derived from satellite measurements of ozone and atmospheric reflectivity.

Results At baseline, the response rate was 70%. At age 10, 90% of those recruited had nevus counts and at age 12, 69%. Children who had migrated to Perth from geographical locations with higher erythemal irradiance had, on average, 1.34 (95% confidence interval, 1.16–1.54) times as many nevi on the back at age 6, 1.29 (1.13–1.49) as many at age 10 and 1.10 (0.92–1.30) times as many at age 12. No significant associations were seen for the face or arms.

Conclusion Ambient solar UV exposure in early childhood is positively associated with number of nevi and nevi develop soon after exposure to sunlight.


NevusPigmentedSunlightUltraviolet raysChildMelanoma


Because melanocytic nevi are strong determinants of their risk of developing melanoma [1], understanding the determinants of nevi should improve our ability to prevent melanoma. It is generally accepted that sun exposure causes nevi to appear during childhood. Ecological studies of nevi have provided some of this evidence [2--5]. For example, in one comparison of children in Brisbane, Australia and Kidderminster, UK, the Australian children had at least five times as many nevi [2].

Ecological studies of nevi involving different observers in the different places are problematic because counting nevi can be subject to substantial inter-observer variation, even when observers have been trained together [6]. Furthermore, counts by the same observer in different locations may vary because they are not blind to this variable.

Ultraviolet (UV) B radiation is the wavelength component of sunlight that causes sunburn (erythema) [7] and is likely to be the component that causes nevi and melanoma [8]. Data on erythemally effective UV irradiance from satellite measurements are available [9] and are preferable to proxies such as latitude for estimating ambient exposure to sunlight. To our knowledge, no study of nevi and only one study of melanoma risk [10] have attempted to use these data.

We report here on the relation between erythemal irradiance and nevus counts in children in a single location, Perth, Australia. The data come from a study of a school-based intervention to reduce sun exposure [11]. Estimates of erythemal irradiance were derived from residential history before entry to the study.

Materials and methods

Study design

The study involved a cohort of children who commenced school in 1995 at age 5 or 6 years [11]. It was a non-randomized trial with schools as the units of intervention. Three groups were included: a control group of 14 schools, a ‘moderate intervention’ group of 11 schools, and a ‘high intervention’ group of eight schools. All children in Grade I in participating schools were recruited, but children of non-European ancestry were not included in analyses.

Ethics committees at The University of Western Australia and Curtin University of Technology approved the study protocol. Parents of all subjects gave written consent.

Ambient sun exposure in the first 6 years of life

In 1995, 1997, 1999 and 2001, parents were mailed questionnaires that asked about their children’s sun-related activities over the previous summer vacation. In the baseline questionnaire, parents were also asked about their child’s skin type, some demographic information, and whether their child had always lived in Perth; if they had not, parents were asked to list the places (town, state, country), and corresponding calendar years where the child had lived for 12 months or more since birth (1989). For 33 children, only the country of residence was given; for 32 who had resided in a small country, the capital city was used, while one child who had resided in the USA was excluded from this analysis. The latitude and longitude of each residence was obtained from Falling Rain Genomics Inc. [12]. Average erythemal irradiance by latitude and longitude for the period 1979--1992 was obtained from the National Center for Atmospheric Research [9]. These data were derived from measurements of ozone and cloud/aerosol reflectivity by instruments on the Nimbus 7 satellite [13]; the erythemal action spectrum [7] was used to calculate the erythemally effective UV irradiance.

The erythemal irradiance for each child, which represents ambient solar UV exposure for the first 6 years of his or her life, was the weighted average of the erythemal irradiance for each place of residence where the weight was the proportion of time (of 6 years) the child resided at that place. For children who had always lived in Perth (approximately 77% of the sample), the value was the erythemal irradiance for Perth (4.226 kJ m-2 day-1).


The primary outcome measure was the number of nevi of any size on the back. Secondary outcome measures were the number of nevi of any size on the face and arms. Nevi were counted in 1995, 1999 and 2001, when the children were 6, 10, and 12 years of age, respectively. Nevi were counted in winter to minimize confusion with freckles and in accordance with a protocol that details methods to distinguish nevi from freckles [14]. Under bright light, lay observers trained by a dermatologist counted nevi on the face and arms. Two observers each examined randomly selected children. After marking anatomic landmarks, slide photographs of the back were taken.

All slides of each child’s back were projected side-by-side onto a whiteboard. A lay observer, trained by a dermatologist, identified and marked all pre-existing nevi on the baseline slide and new nevi on the age ten and twelve slides. The shoulders, which are often freckled, were excluded. All observers were blind to the children’s residential history. To permit estimation of inter-rater reliability for nevi on the back, the dermatologist counted nevi from 47 randomly selected triplets of slides.

Statistical analysis

Because the number of nevi present on each body site was log-normally distributed, it was log transformed after a constant of one was added to allow for subjects with no nevi. The weighted average erythemal irradiance was split into four categories: <3.5, 3.5--4.225, 4.226, and >4.226 kJ m-2 day-1. Children who had lived only in Perth formed the referent group.

The associations between erythemal irradiance and the outcomes, the number of nevi on the back, face and arms, were investigated using linear mixed effects modeling with a random effect for school. Analyses were adjusted for study group (control, moderate intervention, high intervention) and individual-level variables: parental education, southern European ethnicity, sex, propensity to sunburn, hair color and inner arm skin reflectance (skin color) [11]. Body surface area was also considered as a potential confounder, but did not alter any of the regression coefficients for erythemal irradiance and was not included in the final analyses. Interactions between erythemal irradiance and sex of the children and between erythemal irradiance and skin color were examined for all outcomes. For tests of trend across successive categories of ambient sun exposure, a pseudo-continuous variable was modeled using the median values of erythemal irradiance for each child within each category of exposure.

Statistical analyses were performed using Stata 8.2 (Stata Corporation, College Station, TX, USA). All tests of statistical significance are two sided; p<0.05 was considered statistically significant for main effects and to guard against type I error due to multiple comparisons, p<0.01 was considered significant for interactions.


Consent was obtained for 1776 (70%) of those invited to participate; 1614 were of European ancestry and had baseline nevus counts for the face and arms. At age 10, 1449 (90%) of these children had nevus counts for these sites and at age 12, 1115 (69%). Nevi on the back were counted only in children who had a baseline slide and at least one other slide (e.g., baseline and age 10 years); 1411 had baseline counts, 1406 had data at age 10 and 1082 had data for age 12. In all years, over 98% of children with outcome data had complete data for confounding variables.

The proportions of children assigned to the categories of erythemal irradiance of <3.500, 3.500--4.225, 4.226, and >4.226 kJ m-2 day-1 in the first 6 years of life were 5, 9, 77, and 8 percent respectively. At ages 6 and 10 years, children with the highest ambient UV exposure had about 30% more nevi on the back than those who had lived all their life in Perth, whereas by age 12, the difference was only 10 percent and not statistically significant (Table 1). At all ages, children in the two lowest categories of erythemal irradiance had no fewer nevi than those who lived all their life in Perth. For the face and arms, there was little variation in number of nevi by erythemal irradiance (Table 1).
Table 1

Association between erythemally effective UV irradiance at places of residence before age 6 years and the number of nevi on the back, face and arms at age 6, 10, and 12 years

Site & age (years)

No. of children

Erythemal irradiance (kJ m-2 day-1)

p trend

Low <3.5 (median=3.010)

Moderate 3.5--4.225 (3.940)

High 4.226 (4.226)

Very high >4.226 (4.525)




1.14 (0.95,1.37)a

1.08 (0.94,1.24)


1.34 (1.16,1.54)




1.02 (0.85,1.22)

1.01 (0.88,1.15)


1.29 (1.13,1.49)




1.11 (0.91,1.35)

1.05 (0.89,1.24)


1.10 (0.92,1.30)





0.94 (0.80,1.10)

0.99 (0.88,1.12)


1.06 (0.93,1.20)




0.99 (0.82,1.19)

1.04 (0.90,1.20)


1.01 (0.87,1.17)




0.91 (0.74,1.11)

0.86 (0.73,1.02)


0.88 (0.74,1.05)





0.97 (0.83,1.14)

1.04 (0.93, 1.18)


1.05 (0.93,1.20)




1.01 (0.85,1.22)

1.07 (0.93,1.22)


1.07 (0.93,1.23)




1.04 (0.86,1.27)

1.04 (0.88,1.22)


1.05 (0.89,1.25)


a Ratio of mean and its 95% confidence interval in parentheses relative to mean in the (‘High’) reference category of 4.226 kJ m-2 day-1 (Perth), adjusted for sex, southern European ethnicity, parental education, propensity to burn, hair color, inner arm skin reflectance. Analyses of nevi at ages ten and twelve also adjusted for study group

None of the interactions between erythemal irradiance and sex or skin type were significant.

The distribution of erythemal irradiance did not vary by the observer of the number of nevi on the face and arms (for 1995: p-value=0.78). Only one person recorded the number of nevi on the back for all children at ages 6, 10, and 12.

The intra-class correlations (ICCs) from analyses comparing the counts of nevi on the back by the dermatologist and the observer were: 1995, 0.84 ( 95% confidence interval 0.75--0.92); 1999, 0.82 (0.73--0.92); 2001, 0.85 (0.77--0.93) for 2001. The ICCs comparing counts of nevi on the face and arms by different observers in each of the years were: 1995, 0.77 (0.71--0.84) face, 0.84 (0.79--0.89) arms; 1999, 0.83 (0.78--0.89) face, 0.93 (0.91--0.96) arms; 2001, 0.79 (0.70--0.88) face and 0.92 (0.88--0.96), arms.


Until age 10, the number of nevi on the back was 30% higher among children who spent the first 6 years of life living in locations with higher erythemal irradiance than in Perth, Western Australia where the Kidskin study was undertaken. The greater number of nevi on the back associated with high ambient exposure persisted for at least 4 years after all children had been living in Perth. No such associations were found for nevus counts on the face or arms.

A strength of our study is the use of erythemal irradiance as the measure of ambient exposure instead of a proxy such as latitude. The satellite-based figures are available for the whole world and show good agreement with surface-based measurements of UV where the latter exist [15]. Although there is some uncertainty regarding the relative importance of UV A and B radiation with respect to their ability to cause melanoma [16], and therefore probably nevi as well, recent animal evidence suggests that UV B plays a substantially greater role [17--19]. Thus, the erythemal irradiance is the most relevant, widely available measure of ambient exposure.

Another strength of the study is its conduct in a single location. Because observers were blind to the children’s residential histories, observer bias is unlikely and the level of inter-observer variability in each year was low. Nevertheless, measurement error may partly explain the lack of association between ambient exposure and nevi on the face and arms. These sites were counted directly rather than from photographs, whereas for the back, the observer had the advantage of being able to compare photographs at different ages. We previously found that the intervention had a beneficial effect for nevi on the back, but not for the exposed sites [20, 21].

Limitations include loss to follow-up and potential confounding by individual sun exposure habits. By age 12, the differences in number of nevi on the back were smaller than at younger ages, but we are unable to distinguish a waning of the effect over time from potential bias due to loss to follow-up. We could not control for possible confounding of erythemal irradiance by individual sun exposure habits because we had no data on sun exposure habits while children were residing at locations other than Perth. We did not control for individual exposure while the children were residing in Perth because there is a strong possibility that the environment influences individual sun exposure habits. Chance is the most likely explanation for the (non-significantly) slightly higher nevus counts on the back associated with residence in areas with lower erythemal irradiance than Perth.

Previous studies of ambient exposure have compared nevus counts from children living in different locations. Most have used different observers in the places being compared, thus making it difficult to separate real differences from observer variation, which can be substantial [6]. In the first such study, 8- and 9-year-old children from Brisbane, Australia had at least five times as many nevi as did children from Kidderminster, England [2]. The difference in erythemal irradiance between Brisbane (4.1 kJ m-2 day-1) and Kidderminster (1.3 kJ m-2 day-1) was greater than the range in our study, which may partly explain the greater variation in nevus counts. Three-year-old children in Townsville, Australia had about four times as many nevi as children in Glasgow, Scotland [5]. A third study of children 13--15 years of age living in Brisbane, Australia, or Glasgow reported smaller relative differences for nevi on the arm (75% for boys and 9% for girls) [3]. Another Australian study of children in Melbourne (3.0 kJ m-2 day-1), Sydney (3.4 kJ m-2 day-1) and Townsville (5.2 kJ m-2 day-1) involved nevus counts made by the same observers, but not blind to the location of residence [4]. Children in Townsville had about 70% more nevi than children in Melbourne. In that study, by age 15 years the mean counts were as high in Melbourne as in Townsville. Why differences in the number of nevi would diminish with age is unknown, especially when studies of adults suggest that the effects of differences in ambient exposure early in life are maintained into adult life [22, 23]. No association between latitude and number of nevi was found in a study conducted in 16 Italian cities [24]. In summary, despite some minor inconsistencies, the evidence indicates that ambient UV radiation during early childhood is associated with number of nevi and that the association emerges early in life.

There is convincing evidence, particularly from studies of migrants, that childhood sun exposure influences risk of melanoma in adulthood [25]. Development of nevi during childhood in response to sun exposure is a likely link between early life exposure to sunlight and later risk of melanoma. These features of the epidemiology of nevi and melanoma point to early childhood as an important time in which to protect children from excessive sun exposure. However, exposure during adult life should not be ignored. Whiteman et al. [25] reviewed the evidence relating individual sun exposure habits at various times of life to risk of melanoma and reported that exposure in adult life was also important.


Donna Cross, Billie Giles-Corti, and Robyn Johnston contributed to the design and conduct of the study.

Copyright information

© Springer-Verlag 2006